Morphology Materializing: Morphological forms and boundaries in spontaneously written language

• Applicants: Professor Dr. Kristian Berg; Professor Dr. Stefan Hartmann
• Subject Area: Individual Linguistics, Historical Linguistics ; Applied Linguistics, Computational Linguistics
• Term: since 2023
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 533131836

Recent work on the role of morphology in spelling has demonstrated that written language and morphology stand in an intricate relationship (see e.g. Aronoff et al., 2016). This makes written language an important testing ground for lexicon-usage-based morphology. Spontaneous handwritten texts offer an exclusive insight into the relationship between morphology and spelling of individual writers: Firstly, they tend to be produced with little conscious planning, and corrections that the writer made will usually be clearly visible as such. This means that factors influencing written language, including morphology, will manifest themselves more directly in handwritten text as compared to typewritten and more carefully edited texts that make up the vast majority of written language corpora. Secondly, previous research gives us good reasons to expect that the material realization of handwritten text can give clues as to how words are processed and represented in the mind.These two aspects lead us directly to the main goals of our project, which addresses the question of how morphology is manifested in writing from a twofold perspective. We address the question how morphological structure influences the occurrence of spelling errors, and what this relationship tells us about the processing of written units; that is the graphemic perspective. We also address the material realization of handwritten language. Given that psychomotoric research points towards a role of morphology in the material realization of written language (Kandel et al., 2008; Nottbusch, 2008), we will focus on the actual shape of letter forms; that is the graphetic perspective (this angle is analogous to a phonetic analysis of spoken language; see e.g. Meletis 2015). For both parts of the project, we draw on an annotated corpus of hand-written school-exit exams compiled by the first applicant. The corpus will be enriched semi-automatically with morphological annotations. In the morpho- graphemic subproject, we focus on one specific type of spelling errors, namely omissions of individual or multiple letters. We use mixed-effects regression modelling to gauge the effect of different factors on the occurrence of spelling errors, including the morphological structure of the word, the letter(s) involved, word frequency, and the immediate context. For the morpho-graphetic subproject, we further enrich the corpus with a semi-automatic annotation of letter shapes, and we use generalized additive models (GAMs) to assess the impact of various pertinent factors, including morphological structure, on the material manifestation of words in writing. Our investigation of variation in writing provides an important testbed for a lexicon-usage-based approach to morphology, as the influence of factors like entrenchment and analogy can be taken into account in a systematic way.

• DFG Programme: Research Grants